Skip to content

Atomistic-scale simulations of the high-temperature chemistry of Si/C/O/H-based polymers and their conversion to Si/C solid materials

Summary

ReaxFF molecular dynamics simulates high-temperature pyrolysis of polydimethylsiloxane (PDMS) and carbon-enriched analogs (PTMS, PETMS, PDETMS) in small periodic cells and a larger PETMS system, focusing on silicon clustering versus small-molecule products at 2000 and 3000 K, RDF-based amorphous structure of large clusters, and supporting validation runs (density, time length, replicate trajectories).

Methods

Primary source: papers/Ul_Hosna_JPC_C_SiCOH_polymers_2026_galley.pdf (JPCC; galley watermark).

Systems. Five dimers per cell in a 25 x 25 x 25 Angstrom^3 cube at initial density 0.1 g/cm^3 (Table 1 gives formula and density per polymer); eight-fold larger PETMS systems at the same density. Additional PDMS validation at experimental-like density 0.96515 g/cm^3, extended 2 ns duration, and four replicate trajectories for PDMS at the dilute density.

Integration and thermostat. Energy minimization; NVT heating/equilibration at 2000 and 3000 K with Berendsen thermostat (weak coupling, damping constant 100 fs) and timestep 0.05 fs. Production MD 1 ns with coordinates saved every 100 steps (as stated).

Force field. ReaxFF Si/C/O/H parametrization of Soria et al. trained to DFT data for alkyl-grafted Si surfaces; energy decomposition follows standard ReaxFF bonded and nonbonded terms (equation (1) in the article, referencing Chenoweth et al. SI).

Analysis. Bond-order cutoff 0.3 for connectivity; large clusters defined as silicon-containing species with mass greater than 50 amu, small molecules below 50 amu; mass ratio of large clusters to small molecules versus time; radial distribution functions for large clusters after pyrolysis and cooling protocols described in the paper.

1 — MD application (atomistic dynamics). Reactive Molecular dynamics (LAMMPS-class ReaxFF implementation in the article): Five dimers per 25×25×25 ų periodic cell at 0.1 g/cm³ (Table 1) and 8× larger PETMS supercell; NVT heating and pyrolysis at 2000 K and 3000 K; 0.05 fs time step; Berendsen thermostat (damping 100 fs); 1 ns production (coordinates every 100 steps); PDMS validation at 0.96515 g/cm³, 2 ns duration, four replicate trajectories at the dilute density. N/A — no NPT barostat; N/A — no static external electric field; N/A — no replica/metadynamics; N/A — isotropic GPa hydrostatic pressure not used in the NVT pyrolysis (constant-volume cells).

2 — Force-field training — N/A (uses Soria-class Si/C/O/H ReaxFF from DFT-trained alkyl-Si data; not a new fit paper in isolation).

3 — Static QM / DFT as standalone study — N/A (DFT only underpins the Soria ReaxFF training).

4 — Review — N/A.

Findings

  • Pyrolysis at 2000 K yields a higher large-cluster-to-small-molecule mass ratio than at 3000 K for the conditions surveyed; PTMS at 2000 K reaches a maximum reported ratio of 5.4 in the abstract.
  • Simulations produce SiC-rich nanoparticles and silica-type clusters plus small-molecule byproducts including hydrocarbons, aldehydes, and CO depending on temperature and precursor.
  • Eight-fold larger PETMS runs show similar qualitative patterns in small-molecule formation and cluster distributions relative to smaller cells.
  • RDFs of large clusters indicate amorphous Si/C/H/O networks without crystallization on the cooling protocol described; clusters described as high-viscosity, glass-like.

  • PDMS time-extension to 2 ns and alternate density conditions support that 1 ns captures the main compositional trends for the dilute systems used, while high-density PDMS shows slower fragmentation and persistent siloxane aggregates as detailed in the Results section.

  • Corpus honesty: Galley Ul_Hosna_JPC_C_SiCOH_polymers_2026_galley.pdf; Berendsen under exothermic reaction (see ## Limitations).

Limitations

Galley PDF; Berendsen thermostat chosen for stability under exothermic chemistry rather than Nose-Hoover; single trajectories for PTMS, PETMS, and PDETMS at main conditions with multi-trajectory statistics for PDMS only.

Relevance to group

Adri van Duin senior author on ReaxFF pyrolysis modeling of organosilicon precursors toward SiC-related ceramics.

Citations and evidence anchors

DOI: 10.1021/acs.jpcc.5c05359.